Parameterized collision region for centralized motion planning of multiagents along specified paths

초록

This paper presents closed-form analytic solutions for collision detection among multiagents traveling along specified paths. Previous solutions for centralized multiagent systems have mainly used iterative computational approaches for collision detection, which impose a heavy computational burden on the systems. In this paper, we formalize a new mathematical approach to overcoming the difficulty on the basis of simple continuous curvature (SCC) path modeling and a collision representation tool, extended collision map (ECM) method. The formulation permits all the potential collisions to be detected, represented, and parameterized with physical and geometric variables. The proposed parameterized collision region (PCR) method is a simple but precise, computationally efficient tool for describing complicated potential collisions with time traveled. Several simulations are presented to validate the proposed approach for use in centralized collision detectors and to compare the results with those of the iterative computational method and the proximity query package (PQP) method that are available.

키워드